Establishment of an AI-supported scoring system for neuroglial cells.

Abstract

The feasibility of a computer-aided scoring system based on artificial intelligence to detect and classify morphological changes in neuroglial cells was assessed in this study. The system was applied to hippocampal organotypic slice cultures (OHC) from 5 to 7 day-old wild-type and TNF-overexpressing mice in order to analyze effects of a proinflammtory stimulus such as TNF. The area fraction of cells, cell number, number of cell processes and area of the cell nucleus were used as target variables. Immunfluorescence labeling was used to visualize neuronal processes (anti-neurofilaments), microglia (anti-Iba1) and astrocytes (anti-GFAP). The analytic system was able to reliably detect differences in the applied target variables such as the increase in neuronal processes over a period of 14 days in both mouse lines. The number of microglial projections and the microglial cell number provided reliable information about activation level. In addition, the area of microglial cell nuclei was suitable for classification of microglia into activity levels. This scoring system was supported by description of morphology, using the automatically created cell masks. Therefore, this scoring system is suitable for morphological description and linking the morphology with the respective cellular activity level employing analyses of large data sets in a short time.